Hugging Face's logo

adithegod
/
prompt_firewall_v2

Safetensors
Model card Files
xet
 Community
prompt_firewall_v2
File size: 1,829 Bytes
51bf803
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
{
  "version": "2.0",
  "created_at": "2026-06-21T11:07:28.861636",
  "architecture": "XGBoost + DeBERTa-v3-base ensemble",
  "embedding": "sentence-transformers/all-MiniLM-L6-v2",
  "transformer": "microsoft/deberta-v3-base",
  "feature_dim": 396,
  "ensemble": {
    "w_xgb": 0.05,
    "w_transformer": 0.95,
    "w_faiss-delta": 0.1,
    "threshold": 0.915
  },
  "xgb": {
    "n_estimators": 300,
    "max_depth": 5,
    "threshold": 0.5499999999999998
  },
  "training_data": {
    "adversarial": 12000,
    "benign": 12000,
    "families": [
      "JAILBREAK_PROMPTS",
      "ADVERSARIAL_ATTACKS",
      "TECHNICAL_ATTACKS",
      "ENHANCED_JAILBREAKS",
      "CRITICAL_SCENARIOS",
      "ADVANCED_TECHNIQUES",
      "EXTREME_SCENARIOS",
      "JAILBREAK_PROMPTS_Aditya",
      "ARMAAN_KEY_ATTACK",
      "MULTILINGUAL_PROMPTS",
      "ARMAAN_MULTILINGUAL_PROMPTS",
      "multi_lang",
      "LANGUAGE_BASED_PROMPTS_JIYA",
      "SKELETON_KEY_ATTACKS",
      "COMBINATION_ATTACKS",
      "REFUSAL_SUPPRESSION_ATTACKS",
      "AFFIRMATIVE_FORCING_ATTACKS",
      "ROLEPLAY_PERSONA_ATTACKS",
      "MANY_SHOT_ATTACKS",
      "BEST_OF_N_SEMANTIC_ATTACKS",
      "JAILBREAK_ATTACKS",
      "PROMPT_INJECTION_ATTACKS",
      "LOGIC_TRAP_ATTACKS",
      "ADVERSARIAL_CREATIVE_ATTACK",
      "CHAIN_OF_THOUGHT_ATTACKS",
      "SOCIAL_ENGINEERING_SIMULATION",
      "AMBIGUOUS_LANGUAGE_SIMULATION",
      "MULTI_TURN_PERSISTENCE_SIMULATION",
      "ADVANCED_HIGH_FIDELITY_SIMULATION",
      "PERSUASIVE_ARGUMENT_SIMULATION",
      "PHILOSOPHICAL_TRAP_ATTACKS",
      "CONSISTENCY_EXPLOIT_ATTACKS",
      "FALSE_PREMISE_ATTACKS",
      "REDUCTIO_ATTACKS",
      "META_REASONING_ATTACKS",
      "OBFUSCATION_ATTACKS",
      "PERSONA_ATTACKS"
    ]
  },
  "test_metrics": {
    "f1": 0.9808,
    "fpr": 0.0096,
    "tpr": 0.9716
  }
}